Introduction

Taxa included in tribe Cypereae are annual or perennial herbs that vary in stature from minute to 5 m tall. Leaves generally have well-developed blades, but are reduced to lobes in some species; there also may be a ligule. Inflorescences are capitate or anthelate. They all have hermaphrodite, trimerous flowers, with each subtended by a papery glume. Glumes are spirally or distichously arranged in the spikelets, apart from some reduced species in which the arrangement is obscure.

Generic classification in tribe Cypereae and subfamily Cyperoideae dates back to Linneaus (1753), who described the genera Scirpus and Cyperus to include all species of Cyperaceae with bisexual flowers, and distinguished by the spiral versus distichous glume arrangement in Scirpus and Cyperus respectively. The broad circumscription of Scirpus, based on common and widespread characters, resulted in a heterogeneous assemblage which was treated by subsequent workers as one genus (e.g. Boeckeler, 1870; Clarke, 1894, 1898, 1902; Hitchcock et al., 1969) or split into a number of smaller genera (e.g. Brown, 1810; Raynal 1973; Wilson, 1981; Goetghebeur, 1998).

Classification of genera into tribes in Cyperoideae has differed widely among authors, depending on which character(s) were emphasized. Therefore, there is a need to revise generic and tribal circumscriptions and especially incorporate new evidence from morphology and DNA sequence data.

Taxonomic History of Tribe Cypereae

Tribal concepts in Cyperoideae have varied over the years. A large number of legitimate tribal names have been published in Cyperoideae, including Cypereae, Scirpeae, Fuireneae, Ficinieae, Schoenoplecteae, Abildgaardieae, Lipocarpheae, and Eleocharideae (Goetghebeur, 1985). Cypereae and Scirpeae have been the most frequently used tribal names (e.g. Haines & Lye, 1983; Bruhl, 1995; Goetghebeur, 1998). The main difference has traditionally been that Scirpeae have spirally arranged glumes whereas in Cypereae glumes are two-ranked (e.g. Lye, 1971). However, this tribal classification has resulted in genera such as Oxycaryum and Isolepis being classified in Scirpeae even though these genera show closer affinity to Cyperus, as pointed out by Raynal (1973).

Embryological data (e.g. Van der Veken, 1965; Haines & Lye, 1971, 1976, 1983; Raynal, 1973, 1977; Wilson, 1981; Goetghebeur, 1996, 1998; Bruhl, 1995) have contributed significantly to generic and tribal circumscription in Cyperoideae. Heterogeneous Scirpus sensu lato has embryo types characteristic of Cyperus, Carex, Bulbostylis, Fimbristylis, Schoenus and Schoenoplectus, whereas Cyperus has mainly the Cyperus-type embryo. Based on the interpretation that several genera could have the same type of embryo but a single genus should have only a single type of embryo, Scirpus sensu lato was split into several genera. Currently (Geotghebeur, 1998), tribes in Cyperoideae are classified to include genera sharing a single embryo type. For example, Cypereae have the Cyperus-type embryo and the similar Ficinia-type, whereas Scirpeae sensu stricto have only the Fimbristylis-type embryo.

Two recent classifications of Cyperoideae, based predominantly on morphological data, have differed in placement of genera in Cypereae. Goetghebeur (1998) classified all taxa characterised by Cyperus-type embryo in Cypereae, whereas Bruhl (1995) placed genera having spiral glume arrangement (i.e. Isolepis, Ficinia, Desmoschoenus, Scirpoides, Kyllingiella, Oxycaryum) in Scirpeae. Hellmuthia, bearing spirally arranged glumes but having an additional pair of scales in flowers subtended by the most proximal glumes, was placed in Scirpeae by Bruhl (1995) but in Chrysitricheae by Goetghebeur (1998). Bruhl (1995) did not recognise the tribes Eleocharideae and Fuireneae (sensu Goetghebeur, 1998), but included these taxa in Scirpeae.

Phylogeny, Character Homology Assessment and Generic Circumscription

Phylogenetic Relationship Based on Molecular Data

Over the last ten years, molecular systematic data have been used in the classification of ranks above family (e.g. APG, 2003) and in supraqeneric classification within Cyperaceae (Muasya et al., 1998, 2000a; Simpson et al., 2007). Studies on Cyperoideae have targeted phylogenetic relationships in Scirpeae (Muasya et al., 2000b, Dhooge et al., 2003) and Abildgaardieae (Ghamkhar et al., 2007), and focussed on genera Eleocharis (Roalson & Friar, 2000; Yano et al., 2004), Isolepis (Muasya et al., 2001a), Cyperus sensu lato (Muasya et al., 2002) and Schoenoplectus (Yano & Hoshino, 2005).

Total DNA was extracted from leaves or culms collected in the field or from herbarium specimens (Table 1). DNA extraction, amplification and sequencing were performed according to published procedures (e.g. Muasya et al., 2001a, 2002), and the resulting sequences aligned manually. We present and discuss here results of a maximum parsimony analysis of representatives of 18 of the 19 genera in Cypereae recognised by us; no material of the monotypic genus Ascopholis was available. The DNA data matrix (rbcL gene, rps16 intron, trnL intron and trnL-F intergenic spacer) comprises 3,721 characters among which 625 are potentially parsimony-informative. The matrix was analysed using the heuristic algorithm in PAUP* (Swofford, 2002), random addition for 10,000 replicates with tree-bisection-reconnection (TBR). Bootstrap analysis was performed for 1,000 replicates under maximum parsimony criterion (random taxon addition, twin replicates, TBR).

Table 1 List of Taxa Sampled with Vouchers and Genbank Accession Numbers
Full size table

The strict consensus tree generated from the maximum parsimony analysis is presented in Fig. 1, with the bootstrap values for the various branches mapped. Cypereae are resolved into the Cyperus and Ficinia clades. The Ficinia clade comprises Scirpoides, Hellmuthia, Isolepis, Ficinia, Desmoschoenus and two Scirpus species (S. falsus and S. ficinioides). The Cyperus clade has Cyperus sensu stricto as the core genus, in which the thirteen derived genera (Alinula, Androtrichum, Ascolepis, Courtoisina, Kyllinga, Kyllingiella, Lipocarpha, Oxycaryum, Pycreus, Queenslandiella, Remirea, Sphaerocyperus, and Volkiella) are embedded.

Fig. 1
figure 1

Maximum parsimony strict consensus tree of Cypereae based on heuristic analysis of plastid DNA sequence data. Cyperus and Ficinia clades are marked by black and grey bars respectively. Bootstrap support values shown as *for 50–74%, ** for 75–89% and *** for 90–100%

Full size image

Assessment of Morphological Character Homology

Using the DNA phylogenetic framework (Fig. 1), we evaluate the homology of key morphological characters used in classification of the Cypereae. The morphological characters are manually plotted on the DNA topology, majority of characters can be unambiguously reconstructed on the phylogeny. Ascopholis, a monotypic genus restricted to India (Goetghebeur, 1998), has not been included in this study due to unavailability of material. Generic status of Ascopholis is not accepted by all, and it has been suggested to be conspecific to the widespread Cyperus mollipes (C. B. Clarke) K. Schum (Govaerts et al., 2007).

Mature Embryo Morphology

Cypereae are characterised by the presence of a Cyperus-type embryo (Van der Veken, 1965; Haines & Lye, 1971, 1976; Raynal 1973; Wilson, 1981; Goetghebeur, 1985). In the Ficinia clade, species of Ficinia have a Ficinia-type embryo which is similar to Cyperus-type, but Isolepis, Hellmuthia and Scirpoides have a typical Cyperus-type embryo (Van der Veken, 1965; Haines & Lye, 1971). The embryo type in Scirpus falsus and S. ficinioides has not been studied, mainly because mature nutlets were not available.

Concepts of mature embryo morphological states are subject to individual interpretation of homology, and it may be difficult to distinguish similar embryo types in some cases. For example, Isolepis humillima, placed in Isolepis due to the presence of spiral glume arrangement, has been interpreted as having an embryo similar to Scirpoides (Wilson, 1981). The phylogenetic position of this taxon in molecular analyses is within Schoenoplectus subgen. Actaeogeton, a group possessing a Schoenoplectus-type embryo. The mature embryo in Cypereae is less complex when compared to state is the sister tribe Fuireneae (Schoenoplectus type), hence our study does not support Juguet's contention (as reported in Raynal, 1973) that the embryogeny of Cypereae is very evolved compared to the rest of the family.

Annual Versus Perennial Life Form

Annual and perennial growth forms are observed among members of tribe Cypereae (Haines & Lye, 1983; Goetghebeur, 1998; Table 2). In the Ficinia clade, an annual life form has evolved only in Isolepis (which also has some perennial species) whereas all other taxa are perennial. In the Cyperus clade, an annual life form is exclusively found in Courtoisina, Queenslandiella and Alinula; a predominantly perennial life form is observed in Oxycaryum, Kyllingiella, Remirea, Sphaerocyperus, Kyllinga and Ascolepis); while both annual and perennial life forms are recorded in Cyperus sensu stricto, Pycreus and Lipocarpha.

Table 2 Summary of Some of the Diagnostic Characters of the Genera in Cypereae
Full size table

Glume Arrangement

Spiral glume arrangement is a plesiomorphic state in Cyperoideae (Muasya et al., 2001b). In Cypereae (Table 2), the Ficinia clade has predominantly a spiral glume arrangement, except in few species of Ficinia (e.g. F. distans and F. angustifolia) and Isolepis (I. levynsiana and I. venustula). In the Cyperus clade, distichous glume arrangement is usual especially in Androtrichum, Cyperus sensu stricto, Courtoisina, Pycreus, Kyllinga, Queenslandiella, Sphaerocyperus, Remirea, and Volkiella. Oxycaryum, Kyllingiella and Alinula have a spiral glume arrangement, while the spikelet is too reduced in Ascolepis and Lipocarpha for interpretation of glume arrangement (Goetghebeur, 1998). Distichous glume arrangement has evolved more than once in Cypereae, occurring in both the Ficinia and Cyperus clades, and is therefore not unique in Cyperus sensu stricto. The unreliability of distichous arrangement as a diagnostic character has been previously shown (e.g. Raynal, 1973), and evident from our study where taxa with the Cyperus-like distichous glume arrangement (e.g. Isolepis levynsiana) are resolved in the Ficinia clade.

Hypogynous Scales

Hypogynous scales, a character considered plesiomorphic in Cyperoideae, are found in Scirpeae, Fuireneae, Eleocharideae, Dulichieae, and Schoeneae but are absent from Abildgaardieae and Cypereae (Goetghebeur, 1998). Scirpus falsus and S. ficinioides, resolved in Cypereae in molecular phylogenetic analyses (Fig. 1), have bristle-like perianth segments. Similar perianth segments, some well developed and others rudimentary, have been observed in Ficinia material (Muasya et al., unpublished results).

Some florets in Hellmuthia have two scales, which have been suggested to be homologous to scales in Mapanioideae (Haines & Lye, 1976; Goetghebeur, 1998). Recent floral ontogenetic studies (Vrijdaghs et al., 2006) have revealed an adaxially situated third scale in some proximal flowers in spikelets of Hellmuthia, and these are interpreted to be perianth segments and not glumes of reduced florets as in Mapanioideae. Hellmuthia is resolved in the DNA phylogeny among the Ficinia clade and closely related to Scirpus falsus and S. ficinioides.

Gynophores

The gynophore in Cypereae, formed by the development of the hypogynous stalk, is characterised by a lobed cup that envelops the basal part of the nutlet (Vrijdaghs et al., 2005). This structure is absent from the rest of Cyperoideae except for Ficinia, in which variation is observed in size and shape of the gynophore. However, some Ficinia species lack a gynophore, while on the other hand some Isolepis species (e.g. I. marginata) have a rudimentary gynophore (Clarke, 1898; Levyns, 1950; Muasya et al., 2000c, 2001a). A gynophore is present in Alinula lipocarphoides, a taxon previously described in Ficinia and later transferred to Alinula (Kükenthal, 1936; Raynal, 1977), here resolved in the Cyperus clade as sister to Lipocarpha.

Kranz Anatomy

As in most angiosperms families, the plesiomorphic photosynthetic system in most of Cyperaceae is C3 type. Multiple origins of Kranz anatomy are recorded in several lineages including Rhynchospora, Eleocharis, Fimbristylis and Cyperus (Raynal, 1973; Estelita, 1993; Goetghebeur, 1998; Soros & Bruhl, 2000; Muasya et al., 2002; Bruhl & Wilson, 2007). Among Cypereae, Kranz anatomy has evolved once among Cyperus clade and is recorded in Fig. 1 between Cyperus cuspidatus to Alinula lipocarphoides. Bruhl & Wilson (2007) erroneously reported Volkiella to be C3, while in the supporting references they show isotopic carbon reading (−13.6) which is typical for C4.

Samples of Alinula paradoxa and Lipocarpha rehmannii, reported to be C3 (Stock et al., 2004), might have been based on wrongly identified material, especially since there are four other records as C4 for L. rehmannii (Bruhl and Wilson, 2007), and recent carbon isotope studies have confirmed other samples of these taxa to be C4 (Muasya, unpublished results).

Inflorescence Morphology

Inflorescence morphology varies greatly in Cypereae. The basic inflorescence has spikelets in a panicle (Raynal, 1971), which is often modified into an anthela or contracted into a capitate head, spike or reduced to a single spikelet (Goetghebeur, 1998). In Cyperus, C3 taxa tend to have the spikelets arranged in digitate clusters, which is one of the few morphological characters to distinguish the C3 and C4 taxa (which are usually spicately arranged), apart from those species that have the inflorescence reduced to a head (Goetghebeur, 1998). Kükenthal (1935–1936) used this (only partly correctly) to subdivide his subgenus ‘Eu-cyperus’, while Raynal (1973) also noted this (as not being a simple dividing character) particularly in discussing the origins of the ‘Mariscus’ group of species.

Spikelets in a majority of Cypereae have many flowers. Several genera (e.g. Lipocarpha, Ascolepis, Alinula) have pseudo-spikelets, in which spikelets are reduced to single flowers (glumes lost) arranged in cones, each single-flower spikelet subtended by a glume-like bract. The resulting cone resembles a spikelet (Haines & Lye, 1983; Goetghebeur & Vorster, 1988) hence the use of the term ‘pseudo-spikelet’.

Elongation of Filaments

Stamen filaments in most members of Cyperoideae are nearly as long as the glumes and inconspicuous after anthesis. Androtrichum trigynum and A. giganteum have filaments strongly elongating after anthesis, giving the inflorescence a cotton-like look. Such elongation of filaments is not observed in any other species in Cyperoideae.

Dispersal unit

Nutlets (also called achenes by some authors, e.g. Goetghebeur, 1998) in members of Cypereae are dispersed singly or together with elongated filaments, one to a few glumes, or parts of the spikelet axis, or even as complete spikelets (Kükenthal, 1935–1936; Raynal, 1973; Haines & Lye, 1983; Goetghebeur, 1998; Table 2). Courtoisina, Queenslandiella, Kyllinga, Remirea, Sphaerocyperus, Lipocarpha, and Ascolepis have spikelets dispersing as intact units, whereas all taxa in the Ficinia clade, Kyllingiella, Pycreus, Oxycaryum, and Remirea have nutlets dispersed singly. Cyperus has nutlets dispersed either singly or as whole spikelets or variants thereof (notably in Cyperus odoratus).

Nutlet Orientation

Two kinds of nutlet orientation are observed in Cypereae (Table 2). Dorsiventral nutlet orientation is the most common and plesiomorphic state (Kükenthal, 1935–1936; Goetghebeur, 1998; Muasya et al., 2001b). Within Cypereae and Cyperaceae, species with distigmatic styles and dorsiventrally compressed nutlets are observed. Only the genera Kyllinga, Pycreus, and Queenslandiella have lateral nutlet orientation with distigmatic styles and laterally compressed nutlets.

Generic Circumscription

Cypereae are defined here as including all taxa sharing the Cyperus-type of embryo. We expand the tribal circumscription to include characters states such as the occasional presence of floral scales and bristle-like perianth segments, observed in the Ficinia clade.

The Ficinia Clade

Taxa in this clade have a predominantly spiral glume arrangement, but note the presence of distichous glume arrangement in Ficinia and Isolepis. All the genera share ficinioid morphology, e.g. tufted perennials, spiral glume arrangement, and have a center of diversity in the Cape floristic region of South Africa (Goetghebeur, 1998; Archer, 1998; Muasya & Simpson, 2002; Muasya, 2005). The individual genera are diagnosed by a combination of several characters (Table 2), the most notable being the presence of a gynophore and ligule in Ficinia (including Desmoschoenus), presence of two or three scales in the lower florets in Hellmuthia, and perennial growth form and spiral glume arrangement in Scirpoides, whereas Isolepis includes predominantly annual species with a spiral glume arrangement. Two annual species (Isolepis leucoloma and I. levynsiana) with distichous glumes previously described in Cyperus have been transferred to Isolepis, based on morphological and molecular data (Archer, 1998; Muasya et al., 2006, 2007).

There is overlap in generic limits between Isolepis and Ficinia as presently recognised, whereas Desmoschoenus is embedded in Ficinia (Fig. 1). An annual species with rudimentary gynophore described as Isolepis (I. marginata) is resolved in DNA analysis as more closely related to Ficinia. Desmoschoenus and Sickmannia (Ficinia radiata), taxa with a gynophore but with additional unique features, have been recognised as distinct from Ficinia. Phylogenetic results presented here (Fig. 1) show that these taxa are embedded in Ficinia, and should be recognized as members of Ficinia. Sickmannia has already been recognised as Ficinia (F. radiata) in recent treatments (Goetghebeur, 1998; Archer, 2000), whereas Desmoschoenus spiralis, a New Zealand endemic growing in the same coastal habitat as Ficinia nodosa, has no name in Ficinia. More studies are in progress to resolve relationships in the Ficinia clade.

Two of the Scirpus species, S. falsus and S. ficinioides from southern Africa, have the gross morphology of the Ficinia clade, including perennial habit, scapose culms, pseudolateral inflorescences, and spiral glumes. Presence of perianth segments has been used to include these taxa in Scirpus (e.g. Kunth, 1837; Clarke, 1898; Gordon-Gray, 1995) even though typical Scirpus has paniculate inflorescences and nodded culms. So far no embryo studies have been done on these taxa, and attempts to locate appropriate material have not been successful as the taxa rarely produce mature nutlets. Phylogenetic studies resolve these taxa as sister to Hellmuthia (Fig. 1), a pattern that suggests evolution from a southern African ancestor, unlike Scirpus, which is Holarctic. A new genus should be erected to include these two taxa, and more studies are in progress to formalise the recognition of this genus.

The Cyperus Clade

Genera in the Cyperus clade are circumscribed by a combination of morphological characters including spikelet morphology, unit of dispersal, and nutlet orientation (Table 2). Although these genera can be grouped into C3 and C4 anatomical types, there are few observable gross morphological characters to separate the species of Cyperus sensu stricto with the two kinds of anatomy.

Among C3 genera, Androtrichum is diagnosed by the presence of elongated stamen filaments that are persistent and dispersed with the nutlets. However, the two taxa, A. giganteum and A. trigynum, are not sister (Fig. 1) and their shared character state, presence of elongated filaments, may be a parallel adaptation to dispersal in swampy coastal dunes. Kyllingiella and Oxycaryum, previously classified in Scirpeae (e.g. Bruhl, 1995), have a spiral glume arrangement unlike C3 species of Cyperus sensu stricto, which have a distichous glume arrangement (Lye, 1971; Haines & Lye, 1978). Courtoisina has similar morphology to C3 species of Cyperus, but the whole spikelet is dispersed intact.

The C4 genera include a number that are monotypic or with few species (i.e. Queenslandiella, Sphaerocyperus, Remirea, Volkiella, and Alinula), which are separated from the larger genera by a combination of characters. Among the clearly recognizable larger genera are Kyllinga, and Pycreus (together with monotypic Queenslandiella), which have laterally flattened nutlets. Alinula, Volkiella, Ascolepis, and Lipocarpha have highly reduced spikelets. The C4 species of Cyperus sensu stricto have spikelets comprising more than one floret and dorsiventrally compressed nutlets.

There are differences in opinion on whether to recognise Cyperus sensu lato, in a very broad sense with a number of subgenera (e.g. subgenus Kyllinga, and C3 and C4 species of Cyperus sensu stricto in different subgenera; e.g. Kükenthal, 1935–1936), or in a narrow sense with various segregate genera (with Cyperus sensu stricto including C3 and C4 species; e.g. Goetghebeur, 1998). Our results show Cyperus sensu stricto to be polyphyletic, and merging all the segregate taxa into broadly circumscribed Cyperus sensu lato and recognizing various segregates as subgenera would make a monophyletic entity. However, this option is not favored because it would result in a big genus (c. 900 species) and reduce taxonomic clarity. Other partial merging of the taxa into Cyperus, recognizing Oxycaryum, Kyllingiella, Sphaerocyperus, Remirea, Lipocarpha, and Ascolepis as distinct, but treating Courtoisina, Kyllinga, Pycreus, Queenslandiella, and Alinula as subgenera of Cyperus (e.g. Haines & Lye, 1983; Lye, 1997) is not supported by this study.

We follow Goetghebeur (1998) in recognizing Cyperus sensu stricto and recognizing the segregate taxa at generic rank (Table 2) pending more intensive phylogenetic studies to get a full resolution of their relationships.

Future Research

Molecular phylogenetic studies have focused more attention on the Ficinia clade (38% sampling) and less on the Cyperus clade (5% sampling), yet Cyperus clade exhibits wide morphological variation. With more intensive molecular phylogenetic studies and more extensive sampling to include the complete diversity of growth form and morphological types, we expect a better understanding of character homology, which will allow better-informed decisions about generic limits.